Induction of Tumor-Specific Cytotoxic T Lymphocytes in Cancer Patients by Autologous Tumor RNA-Transfected Dendritic Cells

Abstract

Objective

To demonstrate the feasibility of inducing tumor antigen-specific immune responses in patients with metastatic cancer using total tumor RNA-loaded dendritic cells (DCs).

Summary Background Data

The authors have shown that DCs transfected with mRNA encoding defined tumor antigens induce tumor antigen-specific T-cell responses in vitro and in vivo. There may be significant advantages to inducing immune responses against the entire repertoire of antigens expressed by a patient’s autologous tumor.

Methods

RNA was extracted from a metastatic colon cancer and used to load autologous DCs. The DCs were coincubated with autologous T cells and the cytolytic activity of the T cells was assessed by the ability to lyse the autologous tumor cells. RNA was then extracted from a metastatic lung cancer and used to load autologous DCs, followed by four injections of the DC vaccine given every 4 weeks. Tumor antigen-specific cytotoxic T lymphocyte activity was then evaluated by testing peripheral blood mononuclear cells for their ability to lyse an antigen-expressing target.

Results

DCs transfected with the total RNA content of autologous tumor cells stimulated antigen-specific T-cell responses that are capable of recognizing and lysing autologous, primary tumor cells in vitro. Tumor-specific immune responses were induced in a patient with a carcinoembryonic antigen-expressing adenocarcinoma after immunization with autologous DCs transfected with total tumor RNA.

Conclusions

DCs transfected with total tumor RNA may represent a method for inducing immune responses against the entire repertoire of tumor antigens of surgically resected malignancies.

Active immunotherapy, intended to induce antigen-specific T-cell responses to tumor-associated or tumor-specific antigens, has reemerged as a concept for the treatment of advanced malignancies, and perhaps as a modality to prevent cancer metastasis or recurrence. ^1–3 Because T cells recognize antigens as peptide epitopes bound in the groove of major histocompatibility complex (MHC) molecules, ⁴ experimental attempts to induce T-cell responses have included immunization with MHC-restricted, tumor antigen-associated peptides, ^5–7 and proteins representing these tumor-associated antigens along with an adjuvant. ^8,9 An exciting approach is the use of a potent cellular adjuvant, the antigen-presenting dendritic cell (DC), loaded with the antigen of interest. ^10–14

DCs have the exceptional ability to stimulate naive or quiescent CD4⁺ and CD8⁺ T-cell responses in vitro and in vivo. ¹⁵ Vaccination of mice with DCs loaded with antigens in the form of peptides or proteins results in tumor-specific cytotoxic T lymphocyte (CTL) responses capable of rejecting implanted tumors. ^16–19 More recently, we have shown the feasibility of loading DCs with antigen by transfecting the DCs with the mRNA encoding that antigen. ^20,21 One advantage of transfecting DCs with RNA encoding antigen is that mRNAs encode multiple epitopes that can bind to many HLA alleles. Hence, RNA-transfected DCs may be used to stimulate T-cell responses in patients without prior knowledge of, or need to determine, their HLA haplotype. Nonetheless, the mRNA encoding a single known antigen does not permit a particularly broad immune response.

We propose that instead of targeting a single defined antigen, we use DCs transfected with the total mRNA content of autologous tumors (total tumor RNA) to induce T-cell responses against multiple antigens, defined and undefined. We have observed that treatment of tumor-bearing animals with DCs pulsed with tumor-derived RNA leads to a significant reduction in lung metastasis. ²⁰ Transfection of human DCs, generated from the peripheral blood mononuclear cells (PBMCs) of both normal blood donors and patients with advanced cancer, with the mRNA encoding the defined tumor antigen carcinoembryonic antigen (CEA) or undefined total tumor RNA isolated from CEA-overexpressing (CEA⁺) tumor cell lines results in stimulation of a potent primary, CEA-specific CD8⁺ CTL response in vitro. ²¹ Nonetheless, important questions remained about the ability of in vitro generated antigen-specific CTL to recognize and lyse autologous tumor cells and the ability of DCs transfected with total tumor RNA to induce CTL responses in cancer patients.

The purpose of this study was to evaluate the hypothesis that human DCs loaded with the total RNA content of autologous tumor cells could stimulate a CTL response capable of lysing the autologous tumor in vitro and could stimulate a tumor-specific immune response in vivo in patients with metastatic malignancies. We show that DCs, generated from the PBMC progenitors of a patient with metastatic colon cancer and a patient with metastatic melanoma, can be transfected with total tumor RNA from the autologous cancer cells and used to stimulate CTL capable of recognizing and lysing the patient’s own cancer cells or targets expressing the patient’s tumor antigens. In addition, immunization of a cancer patient with DCs transfected with the total tumor RNA derived from his CEA-expressing tumor resulted in tumor antigen-specific CTL responses in vivo. These results provide evidence that T-cell responses specific for defined and undefined tumor antigens can be generated in patients with advanced cancer with total tumor RNA-transfected DCs. This may provide a method for using autologous tumor resected at the time of definitive oncologic surgery as part of an immunization strategy to reduce the risk of progression or recurrence.

METHODS

Patients

Patient 1 was a 62-year-old man with metastatic adenocarcinoma of the colon who underwent an attempted curative resection of recurrent tumor in the retroperitoneal lymph nodes. Patient 2 was a 65-year-old man with a metastatic melanoma involving the liver who underwent a tumor debulking procedure. Patient 3 was a 64-year-old man who presented with a metastatic adenocarcinoma in the subcutaneous tissue and a pulmonary nodule. He refused to undergo chemotherapy, radiotherapy, or surgery of the pulmonary nodule. Two years later, because of enlargement of the nodule, he enrolled in a phase 1 study of active immunotherapy with CEA RNA-transfected DCs. One year later, after the appearance of mediastinal lymphadenopathy, he agreed to undergo a staging mediastinoscopy, at which time tumor involving several lymph nodes was removed.

Cell Lines

An autologous tumor cell line was established from the resected retroperitoneal lymph node metastasis of a colon adenocarcinoma of patient 1. Tumor tissue was mechanically disrupted and propagated in Matrigel basement membrane matrix (Becton Dickinson, Bedford, MA) in RPMI supplemented with 10% fetal calf serum (FCS), 2 mmol/L l-glutamine, 100 μg/mL streptomycin, and 100 U/mL penicillin (complete RPMI). Cells propagated as nonadherent clusters with looped structure. Glandular differentiation was confirmed with hematoxylin-and-eosin staining of a frozen section of a cluster. Clusters were eventually transferred to medium without Matrigel to form a suspension cell line. Immunohistochemical analysis determined that the tumor cells were PAS⁺, CEA⁺, A33⁺, Ki-67⁺, PCNA⁺ (proliferating cell nuclear antigen), and p53⁺. Autologous tumor cells were maintained in DMEM-F12 supplemented with 10% FCS, 25 mmol/L HEPES, 2 mmol/L l-glutamine, 1 μg/mL insulin, and 1 mmol/L sodium pyruvate. A tumor cell line was not established from the tumor tissue of patient 2 or 3. An Epstein-Barr virus (EBV)-transformed cell line was derived from the peripheral blood mononuclear cells of a normal, HLA-A2⁺ individual after in vitro infection with EBV and the addition of cyclosporine A. ²² As a control, the HLA-A2⁺ cervical cancer-derived cell line CaSki (ATCC, Rockville, MD) was maintained in RPMI 1640 supplemented with 10% FCS, 20 mmol/L HEPES, 2 mmol/L l-glutamine, and 1 mmol/L sodium pyruvate. The HLA-A2⁺ CEA-negative adenocarcinoma cell line KLEB was also used as a control in one set of experiments.

Cell lines could not be produced as a result of limiting amounts of tumor and failure of tumor cell propagation in patients 2 and 3.

Isolation of Total Cellular RNA From Autologous Tumor

Total RNA was isolated from the autologous tumor cells of patients 1, 2, and 3, BLCL, CaSki cells, KLEB cells, and autologous PBMCs using RNeasy RNA isolation kits (Qiagen, Santa Clara, CA) according to the manufacturer’s protocol.

Production of In Vitro Transcribed RNA

CEA RNA was made from the template pGEM4Z/CEA/A64. Linearization of the template with Spe1 followed by in vitro transcription with T7 RNA polymerase (Ambion mMessage mMachine, Austin, TX) yields a transcript that contains 47 nucleotides of vector-derived sequence, 85 nucleotides of CEA 5′ untranslated region, 2,106 nucleotides corresponding to the coding region of CEA, 265 nt of CEA 3′ untranslated region, 28 nucleotides of polylinker sequence, 64 adenine residues, and 4 nucleotides from the Spe1 site. Green fluorescent protein (GFP) RNA was made from the template pGEM4Z/GFP/A64. Linearization of the template with Spe1 followed by in vitro transcription with T7 RNA polymerase (Ambion mMessage mMachine) yields a transcript containing 61 nucleotides of vector-derived sequence, the GFP coding sequence, 24 nucleotides of polylinker sequence, 64 adenine residues, and 4 nucleotides from the Spe1 site.

Precursor-Derived Dendritic Cells

Patients and volunteers provided signed informed consent that fulfilled Duke University Medical Center Institutional Review Board guidelines. A leukapheresis was performed on the volunteers and patients and PBMCs were isolated over a Ficoll Hypaque density gradient. DCs were generated from PBMCs in the serum-free medium AIM-V (GIBCO, Grand Island, NY) supplemented with rhGM-CSF and rhIL-4 (gifts of Dr. Mary-Ellen Ryback, Schering-Plough, Kenilworth, NJ) as described, ²³ with minor modifications. ²⁴ PBMCs to be used as responders were cryopreserved at 5 × 10⁷/mL in 90% autologous plasma + 10% DMSO.

Transfecting Dendritic Cells With RNA and Cryopreservation

DCs (5 × 10⁶ cells/mL) in AIM V medium were transfected with CEA RNA (10 μg) or autologous tumor RNA (50 μg). DCs were cryopreserved in 90% autologous plasma + 10% DMSO at 5 × 10⁶/mL. DCs to be used as targets were transfected with RNA in the presence of lipid. RNA, in 250 μL Opti-MEM (GIBCO, Grand Island, NY), and the lipid DMRIE (Vical, San Diego, CA), in 250 μL Opti-MEM, were mixed in polystyrene tubes at room temperature for 5 to 10 minutes. ²¹ The amount of in vitro transcribed RNA used was 1 μg, and the amount of total RNA used was 5 μg per 10⁶ DCs. The RNA to lipid ratio was 1:3. The complex was added to DCs (5 × 10⁶ cells/mL) in Opti-MEM and incubated at 37°C for 20 to 30 minutes. DCs were washed and cultured overnight at 37°C in the presence of GM-CSF and IL-4 before use as targets.

Induction of Primary Cytotoxic T Lymphocyte Responses in Vitro

Stimulation of PBMCs and expansion of CTL was done as described. ²¹ Briefly, autologous PBMCs obtained before any immunization were cocultured with antigen-pulsed DCs in the presence of IL-7 and IL-2. After 10 days, CD8+ T cells were isolated using CD8 microbeads (Miltenyi Biotech, Sunnyvale, CA) per the manufacturer’s protocol. The purity of CD8⁺ T cells was 90% or more by flourescence activated cell sorting (FACS) analysis (data not shown). The captured CD8⁺ T cells were cultured in RPMI 1640 and 10% FCS and 20 U/mL of IL-2 at 37°C. Two days after purification, T-cell blasts were restimulated with antigen-pulsed DCs. ²⁵ CTL assays were done 5 days after restimulation. A standard europium release CTL assay was performed and europium release was measured by time resolved fluorescence. ²⁶ Routinely, cytotoxic activity of CD8⁺ T cells was determined 5 days after the second in vitro stimulation. Alternatively, when the PBMCs were obtained after immunization, nonadherent cells (T and B cells) were used as responders and the cytotoxic activity of the T cells was determined 10 days after a single in vitro stimulation. In some instances, to determine whether CTL activity was generated in the cancer patient, the PBMCs obtained after immunization were thawed, the nonadherent cells were isolated, and CTL activity was assayed 24 hours after thawing with no in vitro stimulation. In all instances T cells were routinely cultured in RPMI 1640 with 10% FCS, supplemented with IL-2 and IL-7.

Targets included autologous tumor cells (when available) and autologous DCs transfected with CEA RNA, GFP RNA (as a control), or total tumor RNA. Specific cytotoxic activity was determined using the formula: % specific release = ([experimental release - spontaneous release]/[total release - spontaneous release]) × 100. Spontaneous release was less than 10% to 12% of total release by detergent in all assays. Standard errors of the means of triplicate cultures were less than 5%.

To test the specificity of immune responses, cold target inhibition was performed by adding back target cells that were not europium-labeled at various ratios to the CTLs while they were being incubated with the europium-labeled target.

Patient Treatment

Patient 3 received an intravenous infusion of 3 × 10⁷ DCs loaded with autologous total tumor RNA over 2 to 3 minutes followed by 1 × 10⁶ autologous total tumor RNA-loaded DCs in a volume of 0.1 mL autologous plasma intradermally into the volar aspect of the forearm or thigh every 4 weeks for four immunizations. Two weeks after the final immunization, a repeat leukapheresis was performed and the presence of cytotoxic T cells specific for the tumor antigen was determined as described above.

RESULTS

Induction of Tumor Antigen-Specific Cytotoxic T Lymphocytes Using B Lymphoblastoid Cells

We wanted to determine whether DCs transfected with total tumor RNA could induce CTLs that recognized and lysed autologous tumor cells. Because it is frequently difficult to propagate resected tumor specimens, we started with a system in which we could generate tumor cells by transforming normal cells in vitro. To this end, B lymphoblastoid cells (BLCLs) were generated from the nonadherent cell fraction of PBMCs as described above. These tumor cells were used as a source of total tumor RNA and eventually as targets to demonstrate tumor-specific CTL activity. As shown in Figure 1A, DCs generated from a normal, HLA-A2 volunteer and transfected with total BLCL RNA induced autologous BLCL-specific CTL activity. Low levels of CTL activity were induced by control RNA (CEA)-transfected DCs and minimal activity was detected against the control target, HLA-A2 cervical cancer cell line, CaSki, which lacks EBV or CEA antigens. Figure 1B shows that the antigen(s) recognized by the CTL induced by DC transfected with BLCL RNA are not present in significant amounts in the B cells from which the BLCL line was derived. The BLCL RNA-transfected DCs stimulate potent autologous BLCL-specific CTL activity but minimal activity against nontransformed B cells. This suggests that autoimmune responses against normal self-antigens are not stimulated by DCs transfected with a pool of RNA that presumably encodes some self-antigens in addition to EBV antigens. Further confirmation of the lack of autoimmune responses is provided by the fact that DCs loaded with autologous B-cell RNA did not stimulate CTLs specific for autologous B cells.

Figure 1. Tumor-specific cytotoxic T lymphocyte (CTL) induction in vitro with dendritic cells (DCs) transfected with total tumor RNA. (A) Induction of CTLs that specifically lyse B lymphoblastoid cells (BLCLs) using autologous DCs transfected with BLCL RNA. RNA, extracted from an EBV-transformed cell line, was used to transfect autologous DCs. Peripheral blood mononuclear cells from an HLA-A2 donor were stimulated in vitro with the transfected DCs as described in text, and the induction of BLCL-specific CTLs was measured using BLCL as targets. The HLA-A2 cell line, CaSki, was a control target. CTLs stimulated by autologous DCs transfected with carcinoembryonic antigen (CEA) RNA was a control. (B) BLCL-specific CTLs show low levels of CTL activity on autologous nontransformed cells. CTLs were induced by DCs transfected with RNA from autologous BLCL, peripheral blood mononuclear cells not adherent to plastic flasks (predominantly T and B cells), T cells, and an allogeneic CEA⁺ tumor cell line. The CTLs were tested for their ability to lyse autologous BLCL or autologous nonadherent cells as a control.

Induction of CEA-Specific and Tumor-Specific Cytotoxic T Lymphocytes With Dendritic Cells Transfected With Total Tumor RNA From Autologous CEA⁺ Tumor Cells

Having shown the generation of tumor-specific CTL activity from cells of a healthy donor using in vitro generated tumor cells as our source of antigen and as the targets, we next attempted to show the in vitro generation of CTLs specific for the autologous tumor of a patient with metastatic (CEA-expressing) colon cancer from which we had established a cell line. The tumor cell line was used for extracting total tumor RNA and eventually as a target to show tumor-specific CTL activity. Autologous DCs were transfected with RNA from a variety of sources: in vitro transcribed CEA RNA; total tumor RNA from the autologous CEA⁺ tumor; control in vitro transcribed RNA encoding an irrelevant protein, GFP ²⁷; or control total cellular RNA from an autologous EBV transformed B-cell line (BLCL) generated from nonadherent cells. The RNA-transfected DCs were used to stimulate patient PBMCs twice in vitro to induce antigen-specific CTLs.

As shown in Figure 2 (left two panels), CEA-specific CTLs (PBMCs stimulated with DCs transfected with in vitro transcribed CEA RNA) were able to recognize and lyse CEA RNA-transfected DCs but not GFP RNA-transfected DCs. Remarkably, tumor-specific CTLs (PBMCs stimulated with DCs transfected with total tumor RNA) were comparable to the CEA-specific CTLs in their ability to lyse CEA-expressing target cells, indicating that the levels of CEA RNA in the total tumor RNA pool were sufficient to stimulate CEA-specific CTLs (notwithstanding the fact that the CEA mRNA species represents a minority of the cellular RNA pool). Stimulation of PBMCs with DCs transfected with GFP RNA induced CTLs capable of lysing GFP RNA-transfected DCs but not CEA RNA-transfected targets.

Figure 2. Induction of carcinoembryonic antigen (CEA)-specific and tumor antigen-specific cytotoxic T lymphocytes (CTLs) in vitro using dendritic cells (DCs) transfected with CEA RNA or tumor RNA from an autologous CEA⁺ tumor. DCs were generated in serum-free AIM-V media and transfected with RNA from in vitro transcribed CEA RNA; control in vitro transcribed RNA encoding green fluorescent protein (GFP); total tumor RNA from autologous CEA⁺ tumor; or control total cellular RNA from Epstein-Barr virus (EBV) transformed B lymphoblastoid cells (BLCL). Peripheral blood mononuclear cells from the patient were stimulated in vitro with the transfected DCs as described in text. Induction of CEA-specific and total tumor RNA-specific CTLs was measured using as targets DCs transfected with CEA RNA or GFP RNA (left two panels) and total tumor RNA or BLCL RNA (right two panels).

Both CEA-specific CTLs and tumor-specific CTLs also lysed DC targets transfected with autologous CEA⁺ tumor RNA but not the control (BLCL) tumor RNA-transfected DCs (see Fig. 2, right two panels). Stimulation of PBMCs with DCs transfected with total RNA from the BLCL (CEA-negative) induced CTLs capable of lysing BLCL RNA-transfected DCs but not autologous tumor RNA-transfected DCs. As predicted, DCs transfected with GFP RNA did not induce tumor-specific CTLs. The levels of lysis obtained with the tumor-specific CTLs were significantly higher than the lysis obtained with CEA-specific CTLs when DCs transfected with total tumor RNA were used as targets. This is consistent with the hypothesis that tumor cells express proteins in addition to CEA that serve as tumor antigens and that unfractionated tumor RNA-transfected DCs elicit responses to a host of other, yet unidentified tumor antigens, whereas in vitro transcribed CEA RNA-transfected DCs stimulate only CEA-specific responses.

We next tested whether the CEA-specific CTLs or the tumor-specific CTLs were capable of recognizing and lysing autologous tumor cells, suggesting that the CTLs generated in vitro by RNA-transfected DCs were capable of recognizing target antigens processed and presented by the tumor cells themselves. Therefore, CEA-specific CTLs and tumor-specific CTLs were used in cytotoxicity assays with autologous tumor cells as targets (Fig. 3). There was recognition and lysis of autologous tumor cells by both CEA-specific and tumor-specific CTLs, but the levels of lysis obtained with tumor-specific CTLs was again significantly higher than the lysis obtained with CEA-specific CTLs. This is consistent with the notion that DCs transfected with total tumor RNA elicit responses to antigens other than CEA. DCs transfected with GFP RNA or total BLCL RNA did not lyse autologous tumor cells.

Figure 3. Induction of tumor-specific cytotoxic T lymphocyte (CTL) response in vitro using dendritic cells (DCs) transfected with carcinoembryonic antigen (CEA) RNA or tumor RNA from an autologous CEA⁺ tumor. DCs were generated in serum-free AIM-V media and transfected with RNA in the form of in vitro transcribed CEA RNA; control in vitro transcribed RNA encoding an irrelevant protein, green fluorescent protein (GFP); total tumor RNA from autologous CEA⁺ tumor; or control total cellular RNA from Epstein-Barr virus (EBV) transformed B lymphoblastoid cells (BLCL). Peripheral blood mononuclear cells from the patient were stimulated in vitro with the transfected DCs as described in text. Induction of CEA-specific CTLs was measured using as a target the autologous colon cancer cell line.

To confirm that tumor-specific CTLs recognized not only CEA but also other, yet unidentified antigens, we performed cold target inhibition assays. As shown in Figure 4, tumor-specific CTLs were generated from PBMCs stimulated with DCs transfected with total tumor RNA. Autologous tumor cells were used as the labeled targets, which were competed out using DCs transfected with autologous tumor RNA, DCs transfected with CEA RNA, or DCs transfected with GFP RNA. DCs transfected with total tumor RNA under nonsaturating conditions of unlabeled to labeled target (6.25:1 and 12.5:1) completely blocked the CTLs from lysing autologous tumor cells. DCs transfected with CEA RNA caused a partial inhibition, and DCs transfected with GFP RNA did not inhibit the lysis of the autologous tumor cells. This experiment, along with data in Figures 2 and 3, shows that total tumor RNA-transfected DCs stimulate CTLs to antigens other than CEA.

Figure 4. Cytotoxic T lymphocyte (CTL) induction by dendritic cells (DCs) transfected with autologous total tumor RNA: cold target inhibition. Peripheral blood mononuclear cells (PBMCs) from the patient were stimulated in vitro with DCs transfected with total tumor RNA from autologous carcinoembryonic antigen (CEA)-positive tumor. Autologous tumor cells were used as the labeled targets and competitors were DCs transfected with autologous tumor RNA, CEA RNA, or green fluorescent protein (GFP) RNA at varying ratios of unlabeled target to labeled target.

Induction of Cytotoxic T Lymphocytes Specific for Antigens From Other Tumor Types

To confirm that tumor antigen-specific activity could be induced using RNA derived from tumors that were not CEA-expressing adenocarcinomas, we transfected DCs with RNA extracted from an hepatic metastasis of malignant melanoma (patient 2) and used them to stimulate autologous PBMCs twice in vitro. Because a cell line could not be established from this tumor and the number of tumor cells was limiting, we tested the resulting CTLs for their ability to lyse targets consisting of autologous DCs loaded with the total tumor RNA. As shown in Figure 5, we were again able to induce antigen-specific T-cell cytolytic activity. The percentage lysis of the melanoma antigen-expressing target was greater than against a target transfected with RNA from a control allogeneic cell line (KLEB).

Figure 5. Induction of tumor-specific cytotoxic T lymphocyte (CTL) response in vitro using dendritic cells (DCs) transfected with autologous melanoma RNA. Peripheral blood mononuclear cells from a patient with melanoma metastatic to the liver were stimulated twice in vitro with DCs transfected with total tumor RNA from the autologous melanoma. Induction of autologous tumor-specific CTLs was measured using as a target autologous DCs loaded with the RNA extracted from the melanoma specimen. Control targets were autologous DC cells loaded with RNA from an allogeneic tumor cell line (KLEB).

Immunization With Total Tumor RNA-Transfected Dendritic Cells Induces Tumor-Specific Cytotoxic T Lymphocytes In Vivo

We tested whether immunization with DCs transfected with total tumor RNA could stimulate tumor-specific T-cell responses in a cancer patient. Patient 3 received four monthly immunizations (intravenously and intradermally) of autologous DCs transfected with RNA extracted from metastatic adenocarcinoma obtained during mediastinoscopy. The immunizations were well tolerated and there were no toxicities observed, but there was no apparent clinical response because the patient continued to have progression of disease.

We tested the CEA and tumor antigen-specific CTL activity in the PBMCs obtained at three different time points: before any immunizations, after immunization with CEA RNA-transfected DCs (on a different clinical trial approximately 6 months before immunization with total tumor RNA-transfected DCs), and after immunization with total tumor RNA-transfected DCs. PBMCs obtained after CEA RNA immunization (i.e., before tumor RNA immunization) were thawed and assayed for their cytotoxic activity before and after a single in vitro stimulation with CEA RNA-transfected DCs. As shown in Figure 6A, induction of CEA-specific CTL activity after immunization with CEA RNA-transfected DCs is evident by the increase in CEA-specific CTLs. Nonetheless, to eliminate the possibility that the in vitro stimulation with CEA RNA-transfected DCs induced the CEA-specific CTLs, we assayed the PBMC samples directly without in vitro stimulation (see Fig. 6B). We again observed CEA-specific CTL activity, albeit very modest, suggesting that the immunizations did induce an increased frequency of CEA-specific precursors.

Figure 6. Demonstration of carcinoembryonic antigen (CEA)-specific cytotoxic T lymphocyte (CTL) activity in the peripheral blood mononuclear cells (PBMCs) obtained after immunization with CEA RNA-transfected dendritic cells (DCs). (A) Cytotoxicity assay after one in vitro stimulation. PBMCs obtained before any immunizations and PBMCs obtained after immunization with CEA RNA-transfected DCs (i.e., before immunization with total tumor RNA-transfected DCs) were tested after one round of in vitro stimulation with CEA RNA-transfected DCs. Targets were CEA RNA-transfected DCs and green fluorescent protein (GFP) RNA-transfected DCs. (B) Cytotoxicity assay with no in vitro stimulation. The two PBMC samples described above were tested directly, without in vitro stimulation, for cytotoxicity against CEA-expressing target cells. Control target was GFP RNA transfected-DCs.

Next we compared the CEA-specific CTL activity and the tumor-specific CTL activity in PBMCs before immunization, PBMCs after CEA RNA DC immunization, and PBMCs after tumor RNA DC immunization. Thus, CEA-specific activity is the cytotoxic T-cell response induced in vivo after immunization with DCs transfected with CEA RNA, and tumor-specific activity is the cytotoxic T-cell response induced in vivo after immunization with tumor RNA-transfected DCs. PBMCs were thawed and assayed for their cytotoxic activity before and after a single in vitro stimulation with total tumor RNA-loaded DCs (Fig. 7). Notably, the lytic activity of the PBMCs obtained after immunization with DCs transfected with total tumor RNA was greater than the PBMCs obtained after immunization with DCs transfected with CEA RNA.

Figure 7. Detection of tumor antigen- and carcinoembryonic antigen (CEA)-specific cytotoxic T lymphocyte (CTL) induction in vivo in a patient immunized with total tumor RNA-transfected dendritic cells (DCs). (A) Cytotoxicity assay after one in vitro stimulation. Peripheral blood mononuclear cells (PBMCs) obtained before any immunizations, PBMCs obtained after immunization with CEA RNA-transfected DCs (i.e., before immunization with total tumor RNA-transfected DCs), and PBMCs obtained after immunization with total tumor RNA-transfected DCs, were tested after one round of in vitro stimulation with total tumor RNA-transfected DCs. Targets were total tumor RNA-transfected DCs and total PBMC RNA-transfected DCs. (B) Cytotoxicity assay with no in vitro stimulation. The three PBMC samples described above were tested directly, without in vitro stimulation, for cytotoxicity against total tumor antigen-expressing target cells. Control target was total PBMC RNA-transfected DCs.

Nonetheless, to eliminate the possibility that the in vitro stimulation with tumor RNA-transfected DCs induced greater activity against tumor RNA-transfected targets, we assayed the three PBMC samples directly, without in vitro stimulation. Once again, the lysis of total tumor RNA-transfected targets by the PBMCs obtained after immunization with total tumor RNA-transfected DCs was greater than the PBMCs obtained after immunization with CEA RNA-transfected DCs. Comparison of Figures 6 and 7 shows that the lytic activity specific for antigens encoded by total tumor RNA does in fact increase after the immunization with total tumor RNA-transfected DCs and suggests that induction of immune responses to antigens other than CEA has occurred in vivo.

DISCUSSION

We have previously shown the induction of primary, CEA-specific CTLs in vitro using human DCs transfected with CEA RNA. ²¹ An ongoing phase 1–2 study is assessing the safety and immunologic and clinical responses after immunization with CEA RNA-transfected DCs in patients with advanced cancers. These studies will address the induction of immune responses to a defined antigen. However, an important issue that remains to be addressed in all cancer immunotherapy strategies is the choice of the antigen: whether it should be a single defined antigen or a mixture of multiple undefined antigens. For example, despite the elucidation of a number of defined tumor antigens from human tumors, a recent study in patients with melanoma revealed that most of the T cells specific for the tumor did not recognize well-known antigens, suggesting that the true tumor rejection antigens may vary from patient to patient or are currently undefined. ²⁸

Clinical studies with autologous tumor cell-BCG vaccines have suggested that immunizing against the entire repertoire of tumor antigens within tumor cells may be an effective method of inducing protective immunity. ²⁹ Unfortunately, it has proven difficult to reliably isolate and expand autologous tumor cells from individual patients in quantities sufficient to generate autologous tumor cell vaccines. DCs are, however, readily generated from PBMCs in the presence of GM-CSF and IL-4. We have previously reported that murine DCs transfected with antigen in the form of total tumor-derived RNA are highly effective in stimulating primary CTL responses in vitro and, more importantly, elicit protective immunity in tumor-bearing animals. ²⁰ One significant advantage of using RNA as the source of the tumor antigen is that small amounts of tumor are necessary to extract sufficient quantities of RNA, and this can be readily amplified using techniques such as polymerase chain reaction.

This report describes the next important step in evaluating total tumor RNA-transfected DCs as an antitumor vaccine: specifically, the ability to transfect DCs with autologous total tumor RNA and induce a tumor-specific immune response in vitro and in vivo. The results show that DCs transfected with RNA (containing CEA RNA) extracted from autologous tumors stimulate CTL activity capable of lysing target cells expressing the defined antigen, CEA, targets expressing the entire range of autologous tumor antigens, and, importantly, autologous tumor cells. Further, the induction of tumor antigen-specific T cells could be accomplished in a different tumor type, not expressing CEA, malignant melanoma.

The levels of lysis obtained with the tumor-specific CTLs (PBMCs stimulated with DCs transfected with total tumor RNA) were higher than the lysis obtained with CEA-specific CTLs (PBMCs stimulated with DCs transfected with CEA RNA) (see Figs. 2 and 3). One explanation for the increased lysis is that the tumor cells express proteins in addition to CEA that serve as tumor antigens, as shown in Figure 4. Total tumor RNA-transfected DCs elicit responses to other, yet unidentified tumor antigens, whereas CEA RNA-transfected DCs stimulate only CEA-specific responses. Therefore, immunization with DC transfected with autologous tumor RNA may provide a strategy to induce tumor-specific T cells from patients against a broad repertoire of tumor antigens. It was particularly interesting to find that a patient with a CEA-expressing malignancy who had previously been immunized with CEA RNA-transfected DCs mounted in vivo immune responses against undefined antigens, which exceeded those against CEA, when immunized with total tumor RNA-transfected DCs (see Figs. 6 and 7).

The results also confirm that total tumor RNA-transfected DCs can be used as targets in cytotoxicity assays to measure the induction of tumor-specific T-cell responses and the lysis of the DC targets is comparable to the lysis of the autologous tumor cells (see Figs. 2 and 3). This would provide a general method for monitoring patients for tumor-specific immune responses that will not depend on propagating the patient’s tumor cells for use as targets. The ability to stimulate and detect tumor-specific T cells using RNA-transfected DCs could provide an assay for detecting novel tumor antigens encoding T-cell epitopes and dramatically enhance the pace of tumor antigen discovery.

One potential drawback of immunizing with unfractionated tumor material, compared with the use of defined tumor antigens, is the increased risk of inducing autoimmune responses with pathologic consequences. There was no evidence of autoimmunity in the cancer patient immunized with total tumor RNA-transfected DCs. Further, when the RNA content of an EBV-transformed tumor cell line derived from autologous PBMCs was used to transfect DCs, the CTLs induced showed minimal lysis of autologous PBMCs (from which the BLCL were derived), suggesting that autoimmunity was not induced (see Fig. 1).

In summary, RNA-transfected DCs are potent antigen presenting cells in vitro and offer a broadly applicable platform for inducing antigen-specific T-cell responses in cancer patients with surgically resected malignancies. These results provide evidence that we can use RNA-transfected DCs to induce tumor-specific T-cell responses in patients with advanced cancer and justify further investigation into this form of cancer immunotherapy in patients with less-advanced disease.

Acknowledgments

The authors thank Doris Coleman, RN, for arranging leukapheresis samples, and Shelley Hull and Eva Fisher for technical assistance.